I was able to reproduce locally over the weekend. I had added an entry for a second IOAPIC on the IOHC, based on the PPRs, but there is not a second device in the MADT. It is looking like that is what is leading to the hang. When i remove the second entry, i'm able to fully boot without amd_iommu=off kernel param. I have a few more experiments to run today, but would like to submit a follow up patch to remove the second IVRS entry. I need to firm up some things from the PPR as well.

View Change

To view, visit change 44897. To unsubscribe, or for help writing mail filters, visit settings.